Chapter 12: OLS Assumptions and Model Building

Links to a carefully curated collection of Youtube videos provide a host of additional sourcs for you to develop your ability to use R software in an engaging and accessible manner.

Video Link 12.1: ‘Regression diagnostics’

Question: What are the key assumptions of the linear regression model identified in the video?

Answer: Linearity between mean response and explanatory variable, normality of errors, constant variance of errors, and independence between observations.


Video Link 12.2: ‘Checking linear regression assumptions in R’

Question 1: Will the OLS assumptions ever be perfectly met when using ‘real’ data?

Answer: No, but we want to assess whether our model does ‘well enough’ in satisfying the assumptions.

Question 2: What are the errors in a regression model?

Answer: They are the difference between the actual y value and the predicted y value for any given value of x.

Question 3: How is non-constant error variance (heteroscedasticity) checked in the video?

Answer: It is checked by examining a residuals vs. fitted values plot. And in the video it appears that the residual are non-constant and thus heteroscedasticity is present.